Scientists Make First Viruses Designed By AI

Scientists have created the first viruses designed by artificial intelligence (AI), marking a significant development in the field with potential benefits and safety concerns. These viruses, known as bacteriophages, target specific types of bacteria and are currently used in medicine to treat stubborn infections. Through the use of genome language models, akin to those behind AI chatbots, researchers led by Dr. Brian Hie from Stanford University were able to design functioning bacteriophage genomes.

These synthetic viruses were subsequently produced in a laboratory and tested against antibiotic-resistant E. coli bacteria, demonstrating their potential to revolutionize phage therapy and expand biotechnological capabilities.

While the prospects of AI-designed viruses offer exciting opportunities, the scientists also highlighted important biosafety, biocontainment, and biosecurity considerations that must be addressed. They emphasized the need for comprehensive safety measures and called for ongoing consultation with experts in safety and security throughout the genome design process. The researchers published their findings in the prestigious journal Science, underscoring the dual-edged nature of this groundbreaking technology.

In response, prominent biosecurity experts, such as Prof. Tom Inglesby and Dr. Moritz Hanke from Johns Hopkins University, underscored the urgent need for robust governance frameworks to manage the implications of such advanced AI applications. They acknowledged the promise of AI-generated viral genomes for scientific advancements but stressed the critical importance of establishing effective safeguards to prevent misuse.

The potential for AI to enable the design of more harmful viruses, even though the complexity of creating complex genomes remains a challenge, prompted calls for controlled access to genetic data and stringent regulations on the creation of potentially dangerous viral genomes.

Moreover, experts like Prof. Tom Ellis from Imperial College London drew attention to the relative ease with which existing pathogens could be manipulated through traditional means, such as "gain-of-function" modifications, as opposed to the complexities involved in AI-generated viral designs. This perspective suggested that while the prospect of AI-designed pathogens warrants careful consideration, the immediate threat it poses might be overstated compared to more straightforward alterations to existing pathogens.

Dr. Filippa Lentzos, a specialist in science and international security at King's College London, advocated for a layered approach to governance, advocating for safeguards not only in the development of AI models but also in the broader context of DNA manufacturing, model development access, responsible research practices, and stringent laboratory biosafety and biosecurity standards.

This comprehensive strategy aims to balance the pursuit of scientific progress with the imperative of maintaining public safety and preventing the unintended consequences of advanced AI technologies.